Cancer National Specialist Advisory Group WALES LUNG CANCER DATA REPORT FOR PATIENTS FIRST SEEN BY LUNG CANCER TEAMS IN 2009 AND 2010

Cancer National Specialist Advisory Group WALES LUNG CANCER DATA REPORT FOR PATIENTS FIRST SEEN BY LUNG CANCER TEAMS IN 2009 AND 2010 Published 2012

Preface The first Wales Lung Cancer Report using prospectively collected data was published in September 2010. This reviewed the clinical data collected on patients with lung cancer and mesothelioma who were first seen in Welsh hospitals in 2007. Interpretation of that first report was hindered by missing data, e.g. data on treatment was available for only 75% of patients. The completeness of the data has improved markedly since then. Whilst there remain concerns about some aspects of the data, the debate can and must focus on the real clinical issues. It is hoped this report, on patients first seen by lung cancer teams and diagnosed with lung cancer in 2009 and 2010, will provide a stimulus to local audit and subsequent improvement in service provision and clinical outcomes for patients with lung cancer in Wales. Acknowledgments This audit report was prepared on behalf of the Cancer National Specialist Advisory Group (NSAG) lung subgroup by: Dr Neil McAndrew, NSAG lung cancer clinical audit lead, Consultant Chest Physician, Wrexham Maelor Hospital Julie Howe, Statistical Analyst, Welsh Cancer Intelligence and Surveillance Unit (WCISU) Rebecca Thomas, Senior Statistician, WCISU Dr Louise Carrington, Programme Co-ordinator, Cancer NSAG Core Team Contributions, advice and comments were gratefully received from: Dr Rachel Butler, Clinical Scientist and Head of Molecular Genetics Laboratory, University Hospital of Wales (UHW), Cardiff Dr Gareth Collier, Consultant Physician, Glangwili Hospital, Carmarthen Dr Jane Hanson, Head of Cancer NSAG Core Team Dr Jason Lester, Consultant Oncologist, Velindre Hospital, Cardiff Dr Ceri White, Principal Statistician, WCISU Dr Ian Williamson, Chair Cancer NSAG Lung Subgroup and Consultant Chest Physician, Royal Gwent Hospital Janet Warlow, Cancer Information Framework Programme Manager Thanks should also go to Debra Bennett, Cancer Information Manager, South Wales Cancer Network and all lung cancer multi-disciplinary teams (MDTs), including data and administration staff, without whom, this report would not have been possible. 2 P age

Contents Preface... 2 Acknowledgments... 2 Executive summary... 4 Recommendations... 5 Introduction... 7 Methods... 7 Summary... 7 General data definitions... 8 Data collection... 8 Analysis... 8 Data presentation... 9 Key clinical indicators... 10 Data quality... 10 Case ascertainment... 11 Completeness of key fields... 12 Clinical outcomes... 14 Lung cancer in general... 14 Histological (and cytological) confirmation of diagnosis... 16 CT scanning prior to bronchoscopy... 16 PET scanning for lung cancer... 18 Percentage of patients discussed at MDT... 19 Percentage of patients seen by CNS.... 19 Active Treatment (specific anti-cancer treatment)... 20 Small Cell Lung Cancer... 21 Non small cell lung cancer (NSCLC)... 23 Histological specification for NSCLC... 24 NSCLC NOS and cytology specimens... 25 PET scanning and NSCLC... 26 Active treatment for all NSCLC... 27 Surgery in NSCLC... 27 Resection rates for all NSCLC... 28 Resection rates for histologically confirmed NSCLC... 28 Chemotherapy rates for NSCLC... 29 Mesothelioma... 31 Incidence by hospital... 31 Chemotherapy for mesothelioma by network... 32 Case mix... 33 EGFR mutation testing... 34 APPENDIX 1: HISTOLOGICAL CATEGORISATION AND ICD 10 CODES... 37 APPENDIX 2: DATA FIELDS AVAILABLE FOR ANALYSIS... 38 3 P age

Executive summary This is the second prospective Wales Lung Cancer Data Report produced using data which has been extracted from the Cancer Network Information System Cymru (Canisc) and deals with patients first seen in 2009 and 2010. The report has focused on two major areas: data completeness and measurements of performance in process and outcome. Data from 4,027 people with a diagnosis of lung cancer were identified for inclusion in this report. Sixteen MDTs were extant in Wales during this period, managing the care of an average 127 new patients a year 1 (range 49-287 2 ). Comparisons made with WCISU registrations demonstrated that patient capture within Canisc at an individual hospital level was generally excellent. Key Clinical Indicators (KCIs) have been agreed by the Cancer NSAG lung subgroup. These are presented by MDT/hospital with reference to data from the National Lung Cancer Audit (NLCA) 3 as a benchmark where possible. Major differences between MDTs in some areas are highlighted within this report. It is acknowledged that different standards of data collection, differences in case mix, and different standards of care between hospitals, may all contribute to the observed variation. It is unlikely that data completeness is a major factor given the marked improvement for the period 2009-2010. Thus the need for individual hospitals to review their pathways for diagnosis, staging and treatment to determine the causes of the variation reported here is clearer than ever before. The key question that arises from this report is how the variation between hospitals in diagnostic process and treatment rates can be explained. In essence, MDTs need to determine the causes of the observed variation and ask whether the poorer performing hospitals are offering a poorer standard of care, or whether their data simply measures a group of patients who were sicker with more advanced disease. The major challenge for the future is to address the clinical challenges within lung cancer services and to support the required changes by providing audit reports that are as timely as possible. The current report, on two years of new patients, is the first step to achieving this. This can only happen if each hospital and network actively addresses their own data in this report and moves to improve either their data collection, or clinical services, or both, as appropriate. Cancer Network lead clinicians need to establish a process for hospitals to share their experience, both with each other, and with overarching organisations involved in clinical and data services. 1 For 2010, mean for 2009 is 124 2 For 2010, range for 2009 is 31 to 262 3 Covers data from the UK in the following cohorts: England and Wales combined, with Scotland and Northern Ireland reported individually. 4 P age

Recommendations It is recommended that LHBs consider the following actions: Each lung cancer MDT and cancer network holds at least an annual meeting to review the results of this report and develop an action plan as necessary. Progress on this will be reviewed by the Cancer NSAG. MDTs work to improve the completeness of data entry into Canisc 4. In light of current performance 5 the 80% target set in 2008 has been revised for these key parameters, but data completeness of other fields needs improvement. For the key parameters of performance status (PS), stage and treatment the aim should be for all three variables to be complete for at least 95% of patients. LHBs, supported by their cancer network, review the number of MDTs they host, considering cross-organisation collaboration if necessary, in order to ensure that key services, such as access to a Clinical Nurse Specialist (CNS), are available to all patients diagnosed with lung cancer. MDTs with outcomes different to the average in any area in this report investigate their data to establish the reasons for these differences and address the issues accordingly. These include: Variation in rates of histological confirmation of lung cancer diagnosis (Figure 5). Low percentages of patients seen by a CNS (Figure 8). High rates of Non-Small Cell Lung Cancer (NSCLC) with sub-type not otherwise specified (NOS) (Figure 14). High and low rates of Positron Emission Tomography Computerised Tomography scans PET for histologically confirmed NSCLC patients (Figure 16). Resection rates for histologically confirmed NSCLC (Figure 19). Chemotherapy rates for patients with stage 3 or 4 NSCLC and a performance status of 0 or 1 (Figure 21). Audit is programmed into the core activities of each lung cancer team, cancer network, and the Lung subgroup of the Cancer NSAG, and takes account of the Cancer Standards and all other extant standard setting work so that each body s audit work is both comprehensive and coherent. 4 Welsh Assembly Government Circular WHC (2008)054 (June 2008) required all cancer MDTs participate in national audits with data completeness at or above 80% 5 For 2009 and 2010 the completeness was 91.2% and 93.7%% respectively. 5 P age

It is recommended that NHS Wales Informatics Service (NWIS), Public Health Wales (PHW) and the Cancer Information Framework (CIF): Further develop the working relationship between Canisc and WCISU teams to enhance cross validation and improve both the quality and the user confidence in the data output of both services. Urgently explore the benefits of all Wales datasets for chemo- and radiotherapy and, along with the other national datasets, develop data linkage as detailed in the Cancer Information Framework (CIF). The Cancer NSAG audit committee will: Broaden the scope to cover more multidisciplinary data Extend the depth of the audit with more detailed data collection on treatment. Enhance the timeliness of reporting, by publishing a summary of findings from the data on patients first seen in 2011 as soon as possible. It is recommended that LHBs, supported by their Cancer Networks, establish a process for hospitals to share their experience both with each other and with overarching organisations involved in clinical and data services, so we can move from data reporting to high quality audit. 6 P age

Introduction The Wales Cancer Standards (2005) 6 require MDTs to ensure all relevant sections of the all Wales Cancer Data Set are completed for each new patient diagnosed with lung cancer, which should enable each MDT to fulfil another of the Standards: The MDT should participate in all Wales clinical audits as specified by the All Wales Lung Cancer Advisory Group. Lung cancer MDTs in Wales have been contributing prospectively collected data to the UK audit (NLCA 7 ) since 2005. This report is the second prospective Wales Lung Cancer Data Report produced using data which has been extracted from Canisc. The first report 8 dealt with the cohort of patients first seen in 2007. This report deals with patients first seen in 2009 and 2010. This cohort was chosen because of both the time frame when the report was written and the completeness of the data for this cohort, which was evidently much improved compared with 2008. This report only deals with patients who presented to secondary care. It does not include patients whose diagnosis and treatment was exclusively in primary care. This report is written for both the clinical and non-clinical reader. It uses data recorded on patients first seen in the four years 2007 2010 to measure data completeness, but the majority of the analyses relate to the data from patients first seen by a lung cancer team in 2009 and 2010. Methods Summary The following steps were undertaken in preparing this report: An audit information pack and an initial patient data extract was sent to each lung MDT lead clinician/coordinator at each LHB/Trust for review and validation. The information pack gave notice that additional data analysis would be undertaken to form a Welsh report. During the validation period clinical teams carried out data verification and added missing information to patient records where appropriate. Clinical sign off from each MDT was received centrally by the lead lung Information Specialist before a final extract of data was taken from Canisc. Death data was sourced from WCISU and was added to the data extract file for the audit submission. Detailed data analysis was performed by WCISU and the lead Information Specialist for lung cancer. 6 Available here: http://www.wales.nhs.uk/sites3/page.cfm?orgid=322&pid=23968 7 Formerly called LUCADA and available here: http://www.ic.nhs.uk/services/national-clinical-audit-support-programmencasp/cancer/lung 8 Available on the Internet: http://www.wales.nhs.uk/sites3/page.cfm?orgid=322&pid=47560, and NHS Wales Intranet: http://howis.wales.nhs.uk/sites3/page.cfm?orgid=322&pid=10629 7 P age

Statistical analysis was carried out by WCISU statisticians in consultation with the audit team. Iterative draft report preparation. Additional data was provided in summary format by the Wales Genetics Service summarising progress in mutational analysis of tissue samples for patients with lung cancer. This was added to the report. Finalised report signed off by lung subgroup of the Cancer NSAG for publication. General data definitions Thoracic malignancy refers to those tumours that have originated within the lungs and pleural cavity, i.e. primary lung cancer and pleural mesothelioma. For the purpose of the presentation of data in this report, primary lung cancer has been further divided into: a) Small Cell Lung Cancer (SCLC) b) Non-Small Cell Lung Cancer (NSCLC) c) Histologically confirmed cases of NSCLC NSCLC is effectively all cases of primary lung cancer other than SCLC. As well as including those cases of NSCLC with histological or cytological proof of that specific cell type, it does, by convention, also include all those cases that have only a clinical and/or radiological basis of diagnosis, i.e. those cases where histology or cytology was not obtained. This is justifiable on the basis that the vast majority of such cases would be expected to be NSCLC, and that the small numbers that were not would make little difference to the analysis. The group of histologically confirmed cases of NSCLC is one which, again by convention, has some particular analyses applied to it, e.g. resection rates (the number of people who have had an operation to remove their cancer). For the ICD 9 and histology codes for the above groups, please see Appendix 1. Data collection As in our previous report, the majority of the data for this publication was extracted from Canisc, the Wales Clinical cancer record/database, having been entered by cancer MDTs as part of routine data capture for patient management purposes. Analysis The analysis was performed by statisticians at WCISU. Throughout the report, many of the charts show error bars, indicating the 95% confidence intervals. 9 International Classification of Diseases 10 th revision 8 P age

A confidence interval quantifies the level of uncertainty in measurement or estimate. A 95% confidence interval is the range of values within which we can be 95% sure that the true value of a statistical measure for the whole population lies. For a number of the data items funnel plots have been developed and included in this report. A funnel plot is designed to give some idea as to whether the outcome for a particular hospital is substantially different from the mean in Wales. For two of the funnel plots the NLCA mean has been used. In the funnel plots each point represents a hospital. The position relative to the x axis reflects the lung cancer caseload for that hospital, and the y axis the parameter under consideration. The dotted lines show 95% (two standard deviations) confidence intervals and dashed lines show 99% (three standard deviations) confidence intervals. If a hospital were to lie outside of the confidence limits then it would be considered to be statistically significantly different from the Wales or NLCA mean with statistically significant higher rates if above the upper dotted line and statistically significant lower rates if below the lower dotted line. Data presentation The table below gives the key to the abbreviations used for individual hospitals. Data is presented by hospital; at the time that this data was collected each hospital ran its own diagnostic service and had its own lung cancer MDT. Thus, each hospital represents an MDT. Patient outcomes are measured against the hospital at which the patient was first seen for their lung cancer, i.e. the hospital to which they were first admitted or seen in outpatients. Ongoing and future re-organisation of services may see a reduction in the number of MDTs in each LHB. Table 1 List of abbreviations used for Welsh hospitals. Hospital/MDT Bronglais General Hospital Glan Clwyd General Hospital Llandough Hospital Morriston Hopsital Neath Port Talbot Hospital Nevill Hall Hospital Prince Charles Hospital Prince Philip Hospital Princess of Wales Hospital Royal Glamorgan Hospital Royal Gwent Hospital Singleton Hospital West Wales General Hospital Withybush General Hospital Ysbyty Gwynedd Ysbyty Maelor Wrexham Abbreviation BGH GCGH LH MH NPTH NHH PCH PPH PWH RGlH RGwH SH WWGH WGH YG YMW 9 P age

References to the NLCA 10 refer to England and Wales data combined. The NLCA report published in 2011 deals with patients with a date first seen (DFS) in 2010 i.e. patients first seen by the lung cancer team in 2010. For clarity, all references to the NLCA are to the DFS of the patient cohort, not the publication date of the report, i.e NLCA DFS 2010. Key clinical indicators One of the functions of the Cancer NSAG is to identify clinical outcomes that indicate the performance of each MDT in key clinical areas. These are called key clinical indicators (KCIs). These priorities will necessarily change as medical pathways evolve and knowledge progresses. Those agreed by the lung cancer sub-group of the Cancer NSAG in 2010 are listed below. KCI 1. Histological / cytological confirmation rate for lung cancer (minimum: 75%, target 85%) KCI 2. Percentage of patients receiving active treatment for lung cancer, additional mean and standard deviations KCI 3. Percentage of lung cancer cases having a resection (minimum: 10%) KCI 4. Percentage of small cell lung cancer patients receiving chemotherapy as first treatment (minimum: 60%) KCI 5. Percentage of non small cell lung cancer patients receiving chemotherapy as first treatment KCIs are reported within this document and are identified as KCI 1-5 where they occur. It has always been intended that these will be updated as changes in clinical practice make them redundant. As a result of clinical advice from the Cancer NSAG, KCIs 3, 4 and 5 have been revised as follows: KCI 3. (revised) Percentage of NSCLC cancer cases having a resection (minimum: 10%) KCI 4. (revised) Percentage of small cell lung cancer patients receiving chemotherapy at any stage (minimum: 60%) KCI 5. (revised) Percentage of non small cell lung cancer patients receiving chemotherapy (at any time, and excluding those receiving surgery). Data quality This section addresses case ascertainment and completeness of key data fields. Have we included all patients with lung cancer who presented to secondary care in the relevant years? Have we got complete data for each of these patients, particularly for the key variables of stage, performance status and treatment? 10 www.ic.nhs.uk/services/national-clinical-audit-support-programme-ncasp/cancer/lung This was previously known as Lucada, and is still often referred to by this former name 10 P age

Case ascertainment For the details of how case ascertainment is assessed (including the apparent contradiction of a case ascertainment of >100%) readers should refer to the detailed explanation given in the 2010 Wales Lung Cancer Report (dealing with patients first seen in 2007) 11. Measured against WCISU records, case ascertainment for individual hospitals is excellent. However, the discrepancy between the secondary care and WCISU totals for Wales as a whole remains a concern even though some of the discrepancy might be accounted for by patients who never came into secondary care, e.g. frail elderly patients diagnosed and treated exclusively by their GP. This is discussed extensively in the previous report. Table 2 Case ascertainment WCISU 2009 Canisc 2009 % case ascertainment WCISU 2010 Canisc 2010 % case ascertainment BGH 32 31 97 44 49 111 GCGH 182 181 99 198 209 106 LH 185 262 142 147 287 195 MH 117 90 77 107 95 89 NPTH 80 89 111 71 92 130 NHH 110 106 96 105 97 92 PCH 123 135 110 94 116 123 PPH 79 96 122 75 95 127 PWH 97 114 118 84 102 121 RGlH 131 139 106 99 131 132 RGwH 196 212 108 174 239 137 SH 123 102 83 98 85 87 WWGH 60 46 77 66 51 77 WGH 65 58 89 88 93 106 YG 142 154 108 114 122 107 YMW 152 146 96 143 160 112 Other welsh 259 362 hospitals English/Extra 144 135 regional hospitals Blank hospitals 124 (111 were DCOs) 165 (138 were DCOs) Total 2401 82 2369 85 DCO = Death Certificate Only, i.e. the only evidence that WCISU have for the diagnosis of lung cancer for these patients is the death certificate. 11 Available on the Internet: http://www.wales.nhs.uk/sites3/page.cfm?orgid=322&pid=47560, and NHS Wales Intranet: http://howis.wales.nhs.uk/sites3/page.cfm?orgid=322&pid=10629 11 P age

Completeness of key fields The charts below (Figure 1 and Figure 2) show the marked improvement in completeness of each of the three individual key parameters of PS, stage and treatment for all Wales between 2007 and 2010. The minimum data completeness has been reported at over 90% for the last two years. This means that those MDTs, with this level of data completeness will be better able to use their case mix adjusted data to understand variations due to sicker patients with more advanced disease, i.e. we have good enough data on each patient to ask whether a hospital which has lower treatment rates does so because they saw frailer patients with more advanced disease compared to other hospitals. The comparable figures for the NLCA report were 85.6% for stage, 84.7% for PS, and 88.7% for treatment. Figure 1. Percentage completeness of staging, performance status and treatment data fields for all lung cancer in Wales, 2007-2010 100 90 80 70 % Complete 60 50 40 30 20 10 0 Staging Performance Status Treatment 2007 2008 2009 2010 80% - 2010 report target The following chart (Figure 2) shows the percentage of individual patient records in which data for all three case mix variables was complete in Wales from 2007 to 2010; this exceeded 90% in 2009 and 2010. The NLCA report does not provide comparable analysis for all three variables however their combined data completeness for PS and stage only is 76.7%. 12 P age

Figure 2. Percentage of individual patient records with complete data for all three key fields (staging, performance status and treatment) for all lung cancer patients Wales, 2007-2010. 100 90 80 70 60 % Complete 50 40 30 20 10 0 2007 2008 2009 2010 Staging, Performance Status and Treatment Complete The chart below (Figure 3) shows the same data as in Figure 2 (i.e. the percentage of individual patient records in which data for all three case mix variables was complete) but by individual hospital. The overall picture is very good, with thirteen out of sixteen hospitals having over 90% completeness in 2010. Some hospitals have improved markedly between 2009 and 2010, e.g. Nevill Hall (NHH) and Neath Port Talbot (NPTH), but others clearly have substantial improvements to make, namely Withybush General Hospital (WGH), Ysbyty Gwynedd (YG) and Princess of Wales Hospital (PWH). Figure 3. Completeness of staging, performance status and treatment data fields in Canisc by hospital, Wales, 2009 and 2010 100 90 Wales = 93.7% Wales = 91.2% 80 70 60 % of patients 50 40 30 20 10 0 BGH GCGH LH MH NPTH NHH PCH PPH PWH RGlH RGwH SH WWGH WGH YG YMW 2009 2010 13 P age

On comparing 2010 with 2009, eleven of the sixteen hospitals showed the same or an increase in completeness for staging, PS and treatment data. A target of 100% completeness is realistic, and some hospitals are already meeting this target. However, technical problems with Canisc can necessitate manual data entry into some fields after the main central programmed extraction of data, (e.g. PS will have to be entered manually if a patient has a data entry of died before treatment in the treatment field). All stage data has been collected using the Tumour Node Metastases system (TNM) version 7 since January 2010. In the NLCA report more than 40% of hospitals in England were still using the old TNM 6 staging system 12. Clinical outcomes This section of the report will be divided into four sub-sections. Whilst there will be a generic section on lung cancer, there will be separate sections for small cell lung cancer, non-small cell lung cancer and mesothelioma, reflecting the fact that whilst there are some similarities in the way the three diseases are investigated and treated, there are also considerable differences. Each section will begin with a short summary of the essential characteristics of the disease and its treatment, thus providing the non-clinical reader with an understanding of the measurements we have chosen to present. Lung cancer in general Lung cancer includes two main groups, NSCLC and the less common SCLC. NSCLC can itself be sub-divided considerably, but the main two cell types are adenocarcinoma and squamous cell carcinoma. These are the important major categories for clinical practice, with relevance for both investigation and treatment. The chart over-leaf (Figure 4) shows the distribution of the different cell types diagnosed in 2009 and 2010. Many audit analyses have historically used all lung cancer as a denominator and should continue to do so, e.g. discussion at MDT, access to CNS. However, some measurements should only be applied to certain sub-groups of patients. For example, SCLC is rarely resected, so including SCLC in the denominator for this analysis is questionable. However, as this is how the analyses have been done historically, some of these figures will still be presented so that comparisons with previous years can be made. Future reports will move away from doing so. 12 National Lung Cancer Audit report, 2011, The NHS Information Centre, available at: http://www.ic.nhs.uk/webfiles/services/ncasp/audits%20and%20reports/nhs_ic_lung_cancer_audit_2011_interactive_pd F_V1.0.pdf 14 P age

Figure 4 Summary of lung cancer sub types diagnosed a) In 2009 b) In 2010 1990 cases 2037 cases 1964 Included in the analysis 26 Removed cases (1 duplicate, 1 non - malignant, 24 anomalies) 2028 Included in the analysis 9 Removed cases (9 anomalies) 70 (4%) Mesothelioma 241 (12%) Small cell lung cancer (SCLC) 1653 (84%) Non small cell lung cancer (NSCLC) 90 (4%) Mesothelioma 256 (13%) Small cell lung cancer (SCLC) 1682 (83%) Non small cell lung cancer (NSCLC) 1065 (64%) Histologically confirmed NSCLC 588 (36%) Non histologically confirmed NSCLC 1080 (64%) Histologically confirmed NSCLC 602 (36%) Non histologically confirmed NSCLC 356 (33%) Squamous cell 709 (67%) Non - squamous cell 371 (34%) Squamous cell 709 (66%) Non - squamous cell 15 P age

Wales lung cancer data report for patients first seen by lung cancer teams in 2009 and 2010: published 2012 Histological (and cytological) confirmation of diagnosis KCI 1 Histological / cytological confirmation rate for lung cancer (minimum: 75%, target 85%) The proportion of patients whose diagnosis has been confirmed with a sample of tissue (either histology or cytology) is regarded as an important measure of the service. Figure 5 shows a mean of two years data for each hospital with the error bars representing the 95% confidence intervals for each hospital. Those hospitals below the NLCA report mean should review their data and diagnostic process. Figure 5 Percentage of all lung cancer patients whose diagnosis is histologically confirmed, Wales (2009-2010 data combined) 100 90 80 70 60 % of patients 50 40 30 20 10 0 NPTH RGlH BGH MH SH WGH YG PPH WWGH YMW PCH NHH GCGH PWH LH RGwH 95% Confidence Intervals for each MDT Wales mean 95% Confidence Interval for Welsh mean England and Wales mean 2010 The mean for Wales, using 2009 and 2010 combined data, is 69% and is shown by the red line. This is lower than that reported in the NLCA report (76%) and is shown by the green line. CT scanning prior to bronchoscopy Almost all patients diagnosed with lung cancer will have a CT scan as part of their assessment. Approximately 50% of patients will have a bronchoscopy. For those patients that do have a bronchoscopy, this should be performed after the CT scan 13 as the CT scan images improve diagnostic performance at bronchoscopy and reduce the number of futile tests performed. The percentage of patients having the CT scan before the bronchoscopy is a measure of the quality of the pathway in a hospital. 13 National Institute of Clinical Excellence Clinical Guideline 21, Lung cancer: The diagnosis and treatment of lung cancer, April 2011 16 P age

Wales lung cancer data report for patients first seen by lung cancer teams in 2009 and 2010: published 2012 The Welsh mean for 2010 is 85.4% (Figure 6) and is similar to that published in the NLCA report at 84.8%. There seems to be a particular problem in Bangor (YG), Withybush (WGH) and Bridgend (PWH). It is worth noting that poor data collection is responsible both for some of the seemingly poor results here, and the apparent improvement between years, e.g. Wrexham (YMW). In Wrexham, data entry for CT scans was very poor in 2009. Hence most bronchoscopies appeared to have been performed without a CT being performed at all, let alone prior to the bronchoscopy 14. Analysis of the data suggests that the increase in performance in 2010 represents better data entry compared with 2009 and not improvement in the pathway. This emphasises the need for good data entry for all fields and not just stage, PS and treatment. However, data completeness is good for the three hospitals with the poorest results (data not presented) and these teams should examine their pathways relating to these investigations 15. This applies particularly to Bridgend (PWH) and Neath (NPTH), where the error bars for both years fall below the respective confidence intervals and indicates consistently lower than average performance. Figure 6 Percentage of all lung cancer patients having a CT scan before bronchoscopy by hospital, Wales, in 2009 and 2010 100 90 80 70 60 % of patients 50 40 30 20 10 0 PPH PCH RGlH RGwH GCGH YMW BGH MH NHH LH SH WWGH WGH YG NPTH PWH 2009 2010 95% Confidence Intervals for each MDT Wales mean 2009 Wales mean 2010 95% Confidence Interval for Welsh Mean 2009 95% Confidence Interval for Welsh Mean 2010 14 Personal communication of local audit 15 The author is aware that the rates in Ysbyty Gwynedd have been reviewed and presented at a North Wales Lung Cancer Network Meeting and the rate in 2011 is substantially improved, another example (as in Wrexham above) of Canisc data being used to review local performance. 17 P age

Wales lung cancer data report for patients first seen by lung cancer teams in 2009 and 2010: published 2012 PET scanning for lung cancer A Positron Emission Tomography Computerised Tomography scan (referred to in this document as a PET scan) is an essential investigation for patients whose performance status and stage of disease at presentation suggest that they may be a candidate for radical treatment, whether resection, radical radiotherapy or chemoradiation. Rates for a hospital may reflect how aggressively MDTs stage disease. The advent of combined modality treatments (chemotherapy and radiotherapy together) for locally advanced disease means that PET scan rates are unlikely to simply reflect rates of presentation of resectable disease. Without controlling for case-mix it is difficult to determine an ideal rate for patients having a PET scan. The mean across 2009-2010 was 18.2% (Figure 7). There is a four fold variation in PET rates between hospitals. It is unlikely that this will be explained simply by variation in case mix (see section on case mix for analysis). The NLCA report has no published data for this. The PET rates are approximately 50% higher than in the first Wales report for patients DFS 2007. However, the PET service then was considerably different, and data quality in many areas less good, so it is difficult to make meaningful comparisons. Figure 7 Percentage of all lung cancer patients having a PET scan by hospital, Wales (2009-2010 data combined) 16 100 90 80 70 60 % of patients 50 40 30 20 10 0 RGlH PCH YMW NHH SH MH YG RGwH LH PPH NPTH WGH BGH GCGH PWH WWGH 95% Confidence Intervals for each MDT Wales mean 95% Confidence Interval for Welsh Mean 16 These rates are for patients proven to have lung cancer and do not include scans performed for other indications or on patients ultimately proven to have benign disease. 18 P age

Percentage of patients discussed at MDT Data for this audit is primarily collected via the MDT meeting so one would expect that rates for this parameter would be high; 98% of patients in this audit were recorded as having their case discussed at an MDT Meeting in 2009, and 96% in 2010. Percentage of patients seen by CNS. The lung cancer CNS plays a pivotal role in the care of patients with lung cancer, both in the diagnostic and treatment parts of the pathway. The percentage of patients who saw a CNS was 76.3% (mean across 2009 and 2010 data), although there is considerable variation between units within this. The NLCA report mean was 75.4%. For some hospitals missing data is the issue, e.g. the Royal Gwent Hospital (RGwH) in 2009 (Table 3), but for other hospitals there seems to be a problem with access to a Lung CNS. A number of hospitals have only a single lung cancer CNS 17. For the audit period only Wrexham (YMW), the Royal Gwent Hospital (RGwH) in Newport and Llandough Hospital (LH) in Cardiff had more than one CNS. The situation has not changed. Analysis from the NLCA report shows that patients seen by a lung cancer CNS were more likely to have active treatment, although no cause and effect relationship has been shown. Figure 8 Percentage of all lung cancer patients seen by a clinical nurse specialist (CNS), by hospital, Wales (2009-2010 data combined) 100 90 80 70 60 % of patients 50 40 30 20 10 0 NPTH NHH GCGH MH SH WGH YMW PCH RGlH PPH LH YG PWH WWGH RGwH BGH 17 personal communication from Wales Lung CNS Group 19 P age

This parameter has always been defined as contact with the lung cancer CNS. However, some patients presenting with advanced disease may be best seen by a palliative care CNS rather than a lung cancer CNS. Thus, for future reports we will additionally look at how many patients were seen by a palliative care CNS, as this will enrich the analysis, and more closely reflect the reality of the pathway. Table 3 Completeness of data by hospital for Seen by CNS field, by hospital, Wales, 2009-2010. 2009 2010 % Seen by CNS % Not seen by CNS % Data missing % Seen by CNS % Not seen by CNS % Data missing BGH 13 70 17 0 100 0 GCGH 92 5 2 91 9 0 LH 62 20 17 81 12 7 MH 91 0 9 92 0 8 NPT 98 2 0 94 6 0 NHH 93 5 2 95 5 0 PCH 92 8 0 82 18 0 PPH 87 8 5 71 19 10 PWH 55 29 16 66 8 26 RGlH 87 12 1 81 11 8 RGH 1 0 99 83 5 13 SH 92 3 5 89 1 10 WWGH 68 20 11 30 36 34 WGH 96 2 2 87 7 7 YG 60 9 30 86 10 3 YMW 88 6 6 92 8 0 Total 71 10 19 81 12 7 Whilst clearly demonstrating the improvement in data capture between 2009 and 2010 Table 3 highlights the need for data completeness in areas other than PS, stage and treatment. Active Treatment (specific anti-cancer treatment) KCI 2 Percentage of patients receiving active treatment for lung cancer Active treatment includes surgery, chemotherapy, radiotherapy and other disease specific treatment at any time, but excludes best supportive care and specialist palliative care. There is marked variation in active treatment rates across Wales (Figure 9), although for Wales as a whole the mean (58.3%) is comparable with the NLCA report mean of 58.4%. The mean 2009-2010 active treatment rates reported by West Wales General (WWG), Prince Philip (PPH), the Royal Gwent 20 P age

(RGwH) and Llandough (LH) are significantly lower than the Wales mean. WWG, PPH and RGwH are also in the lowest quartile reported by the NLCA. Figure 9 Active treatment rates for all lung cancer patients by hospital, Wales (2009-2010 data combined) 70 RGlH NPTH 65 60 BGH WGH MH NHH PCH PWH YG YMWH GCGH % of patients 55 50 SH PPH RGwH LH 45 WWG 40 0 100 200 300 400 500 Number of patients Data Wales Mean 2SD limits 3SD limits Small Cell Lung Cancer SCLC is a rapidly progressive tumour which is often locally advanced at the time of presentation. Furthermore, clinical experience has shown that occult micro-metastases are usually present at the time of diagnosis. In 2010, only 4% of patients with SCLC in Wales had stage I or II disease, compared with 30% of patients with NSCLC. For these two reasons, surgery is usually inappropriate for this type of tumour. The treatment of choice is chemotherapy with or without concurrent or subsequent radiotherapy. Chemotherapy can be of considerable benefit in patients with poorer PS, which is different to NSCLC, so overall one would expect a majority of patients to receive such treatment. The chart overleaf (Figure 10) shows that approximately 80% of patients diagnosed with SCLC are assessed as PS 0 2. This supports the assertion that chemotherapy rates in excess of 65% should be attainable. 21 P age

Figure 10 Percentage of all SCLC patients that have a performance status, Wales, 2009 and 2010 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2009 2010 PS 0 and PS 1 PS 2 PS 3 and PS 4 The existing KCI dealing with chemotherapy was felt by the audit group to need refining and aligning with that used in the NLCA. The lung subgroup of the Cancer NSAG has agreed the following revision: KCI 4 (revised) Percentage of small cell lung cancer patients receiving chemotherapy at any stage (minimum: 60%) Figure 11 shows the chemotherapy rates for SCLC for each hospital in 2009-2010 and includes patients who have had chemotherapy at any point in their pathway. Results are comparable with those in England: the NLCA report mean was 65.1%. Figure 11 Percentage of SCLC patients who have received chemotherapy, by hospital, Wales (2009-2010 data combined) 100 90 80 70 60 % of patients 50 40 30 20 10 0 BGH MH WGH LH PCH NPTH NHH RGlH PPH YG GCGH WWGH YMW RGwH SH PWH 95% Confidence Intervals for each MDT Wales mean 95% Confidence Interval for Welsh Mean 22 P age

Wales lung cancer data report for patients first seen by lung cancer teams in 2009 and 2010: published 2012 The rapidly progressive nature of SCLC means that chemotherapy should be started promptly. Figure 12 below shows the percentage of patients with SCLC receiving chemotherapy who received that treatment within ten days of the decision to give that treatment. More than three quarters of patients who receive chemotherapy for SCLC commence that treatment within ten days of the decision to treat being made. This is encouraging, but there is clearly room for improvement and performance in Princess of Wales (PWH) and the Royal Glamorgan (RGlH) Hospitals was relatively poor in comparison to the best performing MDTs. Figure 12 Percentage of patients with SCLC commencing chemotherapy within 10 days of the decision to treat, by hospital, Wales (2009-2010 data combined) 100 90 80 70 60 % of patients 50 40 30 20 10 0 NHH YMW SH GCGH PCH WWGH MH YG LH BGH PPH WGH RGwH NPTH RGlH PWH 95% Confidence Intervals for each MDT Wales mean 95% Confidence Interval for Welsh Mean Non small cell lung cancer (NSCLC) NSCLC is a more slowly progressive tumour than SCLC and whilst at presentation there is often locally advanced disease (stage III) or distally advanced/metastatic disease (stage IV), there are significant numbers of patients with more limited disease (stage I and II). Furthermore, occult metastases are less frequently present at diagnosis. Broadly speaking, stage III and IV patients may be considered for palliative treatment with chemotherapy and radiotherapy, singly or in combination, whilst the patients with stage I and II disease may be considered for curative (radical) treatment, either with surgical resection or with radical radiotherapy. However, it is worth noting that some Stage III patients, perhaps as many as 30%, may be suitable for radical treatment with chemoradiation or even surgery. Some patients are not fit enough for treatment regardless of stage because of other diseases or co-morbidities (performance status 2, 3 and 4). Patients with such co-morbidities may actually do worse with such treatment than with best supportive care. Thus, many treatments are 23 P age

limited to patients with good performance status (0 or 1), and as Figure 13 shows, this is just over 50% of patients with NSCLC. Figure 13 Percentage of all NSCLC patients that have a performance status, Wales, 2009 and 2010 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2009 2010 PS 0 and PS 1 PS 2 PS 3 and PS 4 Histological specification for NSCLC Drug therapy for NSCLC will differ between squamous cell carcinoma and non-squamous NSCLC (which are mainly adenocarcinoma) so accurate histological and cytological classification of samples is important. Samples in which classification beyond the broad sub-type of NSCLC was not possible or not attempted are described as NSCLC Not Otherwise Specified (or NSCLC-NOS). The rate at which pathology laboratories make this distinction is presented below for both histology and cytology specimens combined, with a low rate being desirable, indicating greater specification of cell type; the percentage figure indicates the proportion remaining NOS after laboratory analysis/reporting. The NSCLC NOS rate is high in some hospitals (Figure 14). The data does not tell us whether this is due to lack of specificity in the reporting of specimens or lack of specificity in the data entry into Canisc, but interrogation of local histopathology records would answer this question very easily. MDTs, particularly those consistently and significantly above the Wales average, are encouraged to work with their pathology departments to reduce their NOS rates. 24 P age

Wales lung cancer data report for patients first seen by lung cancer teams in 2009 and 2010: published 2012 Figure 14 NSCLC- NOS rates by hospital, Wales, 2009 and 2010 100 90 80 70 60 % of patients 50 40 30 20 10 0 RGlH PWH GCGH MH NPTH LH NHH PCH RGwH SH YG PPH WGH WWGH YMW BGH 2009 2010 95% Confidence Intervals for each MDT Wales mean 2009 Wales mean 2010 95% Confidence Interval for Welsh Mean 2009 95% Confidence Interval for Welsh Mean 2010 NSCLC NOS and cytology specimens Increasingly, the tissue samples taken from patients are small cellular samples requiring cytological analysis rather than bigger biopsies requiring histological analysis. As in Figure 14 above, a low rate here (Figure 15) is desirable. Figure 15 Percentage of NSCLC NOS patients with cytological basis of diagnosis, by hospital, Wales (2009-2010 data combined) 100 90 80 70 60 % of patients 50 40 30 20 10 0 NHH RGlH GCGH PWH RGwH MH BGH YG NPTH WGH SH YMW LH PCH WWGH PPH 95% Confidence Intervals for each MDT Wales mean 95% Confidence Interval for Welsh Mean Ability, training and enthusiasm for cytology work amongst histopathologists is variable and not all units will have access to dedicated cytopathologists. LHBs should determine whether this is due to 25 P age

Wales lung cancer data report for patients first seen by lung cancer teams in 2009 and 2010: published 2012 data entry, quality of samples provided to histology or access to specialist cytological opinion. Those above the average (Nevill Hall Hospital [NHH] and Royal Glamorgan Hospital [RGlH]) should investigate their data. PET scanning and NSCLC A PET scan is an important staging investigation for patients with NSCLC and is usually performed in patients who may be candidates for radical treatment. The low rates of use of this investigation (Figure 16) may reflect local casemix or difficulty of access with PET, but could also reflect suboptimal or less aggressive staging pathways in a hospital. Figure 16 Patients with a histologically confirmed NSCLC diagnosis who received a PET scan, by hospital, Wales, (2009-2010 data combined) 100 90 80 70 60 % of patients 50 40 30 20 10 0 RGlH YMW PCH NHH SH MH RGwH PPH LH YG WGH NPTH BGH GCGH PWH WWGH 95% Confidence Intervals for each MDT Wales mean 95% Confidence Interval for Welsh Mean There is nearly a four fold variation in PET rates between hospitals. This is unlikely to be explained by a four fold variation in relevant casemix and raises concerns about assessment of stage in some hospitals. Those MDTs that are significantly above (Royal Glamorgan [RGlH] and Wrexham [YMW]) and below the Welsh average (Glan Clwyd [GCGH], Princess of Wales [PWH] and West Wales General [WWG]) should be subject to further examination. The NLCA do not report on this measurement. 26 P age

Active treatment for all NSCLC Active treatment rates in the NLCA report are expressed as a percentage of all lung cancer cases. Expressing them separately for SCLC and NSCLC is preferable. Active treatment rates for NSCLC are presented below (Figure 17) in a funnel plot. The active treatment rates in four hospitals in Wales, (West Wales General [WWG], Prince Philip Hospital [PPH], the Royal Gwent [RGwH] and Llandough [LH]) are significantly lower than the mean for Wales. Figure 17 Active treatment rates for all NSCLC by hospital, Wales, (2009-2010 data combined) in Wales, 2009-2010 RGlH 65 NPTH 60 WGH NHH MH PCH PWH YG YMWH % of patients 55 50 BGH SH GCGH RGwH LH 45 PPH WWG 40 0 100 200 300 400 Number of patients Data Wales Mean 2SD limits 3SD limits The NLCA do not report on this measurement. Surgery in NSCLC Resection (removal) of a tumour is usually performed only for NSCLC. Traditionally it has been expressed as a percentage of all NSCLC cases or as a percentage of those NSCLC cases with histological confirmation of disease. Both are shown overleaf. The data is presented with the NCLA mean to highlight the low resection rates in Wales. 27 P age

Resection rates for all NSCLC KCI 3 Percentage of NSCLC cancer cases having a resection (minimum: 10%) 18 The mean surgical rate for Wales for all NSCLC in 2010 was 9.8% with the NLCA reporting 13.7%. The resection data (2010) for the majority of Wales hospitals falls in the lowest quartile for NLCA data reported for the same year. Combining the data for 2009 and 2010, still shows that the resection rates in many hospitals in Wales are significantly lower than the mean reported in the NLCA report (Figure 18). Bronglais General Hospital (BGH), Withybush General Hospital (WGH), Neville Hall Hospital (NHH), Prince Philip Hospital (PPH) and Ysbyty Gwynedd (YG) are more that 2 standard deviations from the mean, and West Wales General (WWG), Princess of Wales Hospital (PWH), Prince Charles Hospital (PCH), Royal Gwent Hospital (RGwH) and Llandough Hospital (LH) are more than 3 standard deviations from the mean. Figure 18 Resection rates for all NSCLC by hospital, Wales, (2009-2010 data combined) using the NLCA report mean 25, 20 15 MH % of patients 10 5 BGH NPTH WGH PPH WWG SH NHH PWH YMWH GCGH RGlH YG PCH RGwH LH 0 0 100 200 300 400 Number of patients Data NLCA England and Wales mean 2010 2SD limits 3SD limits Resection rates for histologically confirmed NSCLC When resection rates are expressed as a percentage of histologically confirmed NSCLC (Figure 19) the findings are very similar to those for all NSCLC in Figure 18. The mean surgical rate for 18 As discussed in the methods, this is a refinement of the key clinical indicator agreed in 2010 and is being taken through the formal process to agree this change. 28 P age

histologically confirmed NSCLC in Wales in 2009 and 2010 was 14.6%, whilst that for the NLCA in 2010 was 18.3%. Slightly fewer hospitals (eight in comparison to ten) are significantly different to the NLCA reported mean than for Figure 18. Resection rates in Wales are in the lower NLCA quartile for the majority of hospitals. Figure 19 Resection rates for histologically confirmed NSCLC by hospital, Wales, (2009-2010 data combined) using the NLCA report mean pat e ts by a es, 009 0 0 30 Resected cases as a percentage of histologically confirmed NSCLC % of patients 25 20 15 10 5 0 MH YMWH SH GCGH NHH NPTH LH YG WGH RGlH RGwH PPH PWH PCH BGH WWG 0 Total 50 number of cases 100 of histologically 150 confirmed NSCLC 200 250 Number of patients Data NLCA England and Wales mean 2010 2SD limits 3SD limits Chemotherapy rates for NSCLC The KCI relating to chemotherapy for NSCLC was originally defined as follows; KCI 3. (revised) Percentage of non small cell lung cancer patients receiving chemotherapy (at any time, and excluding those receiving surgery). A two-fold variation is seen in the percentage of patients, excluding those treated surgically, receiving chemotherapy (Figure 20). This indicator does not adjust for case-mix and as the use of chemotherapy in NSCLC is largely limited to patients with stage III and IV disease with PS 0 and 1, data for that patient cohort is presented overleaf. 29 P age

Wales lung cancer data report for patients first seen by lung cancer teams in 2009 and 2010: published 2012 Figure 20 Chemotherapy rates for NSCLC patients, excluding surgical patients, by hospital, Wales (2009-2010 data combined) 100 90 80 70 60 % of patients 50 40 30 20 10 0 BGH WGH NPTH YG SH YMW GCGH MH NHH RGwH PCH PPH WWG LH PWH RGlH 95% Confidence Intervals for each MDT Wales mean 95% Confidence Interval for Welsh Mean As described at the beginning of the NSCLC section, patients with locally advanced (Stages IIIA and IIIB) and metastatic (stage IV) NSCLC would be treated with chemotherapy if they are fit enough (PS 0 & 1 patients mainly). Figure 21 Chemotherapy rates for patients with stage 3 or 4 NSCLC and a performance status of 0 or 1, by hospital, Wales (2009-2010 data combined). 100 90 80 70 60 % of patients 50 40 30 20 10 0 BGH YMW WGH SH NPTH NHH YG GCGH RGwH MH PWH PCH WWGH PPH LH RGlH 95% Confidence Intervals for each MDT Wales mean 95% Confidence Interval for Welsh Mean 30 P age

Leaving aside Bronglais Hospital with its low number of patients, there remains up to a two fold variation in chemotherapy rates for this subgroup of patients between Welsh hospitals. Whilst variation in assessment of staging and PS may explain some of these differences, it is unlikely to be the whole explanation. Further work is required to explain these differences. The NLCA did not report data comparable with Figure 21, although for Stage IIIB and IV combined the NLCA mean was 52.8%. The NLCA report excluded stage IIIA patients from their analysis, perhaps because some (a minority) would be treated with radiotherapy and chemotherapy rather than chemotherapy alone. Mesothelioma Mesothelioma is a malignant tumour that can occur both in the pleura (the lining of the lung and rib cage) and the peritoneum (the lining of the abdomen). This report is concerned with pleural mesothelioma only. Mesothelioma tends to spread along the inside of the rib cage and outer surface of the lung. It is frequently complicated by the accumulation of pleural fluid around the lung causing breathlessness a pleural effusion. There is no proven radical or curative treatment for pleural mesothelioma. Extra pleural pneumonectomy (EPP) is a major operation for mesothelioma with unproven radical intent, and it is not widely practiced. Indeed, a recent study suggested that EPP was actually detrimental compared to no surgery. Other surgical procedures for mesothelioma include pleurectomy and decortication, both essentially debulking procedures, again with limited evidence base but more widely practiced than EPP. All of these procedures are still being evaluated in clinical trials. Pleurodesis, a palliative procedure designed to prevent recurrence of a pleural effusion can be performed surgically or nonsurgically. Chemotherapy has an established palliative role in pleural mesothelioma. The lack of evidence for the surgical procedures in mesothelioma, and the small number of cases for each unit meant that we have not published any surgical data. Incidence by hospital As can be seen from Table 4 (overleaf), incidence of mesothelioma is low with most hospitals seeing fewer than 10 cases per year. 31 P age